There is no server to switch off.
This page, the articles on it and the archive behind them are served by independent machines run by people, not by a data centre we rent. That is the whole promise of the platform, and it is only true for as long as enough of you run one.
One article, requested from four places.
A reader asks for a piece. The request goes to whichever node is closest to them, not to a single origin server somewhere else in the world. Every node holding a copy can answer, and the copy is checked against the fingerprint recorded when the piece was published, so nobody can serve you an altered version.
An illustration of the delivery model, not live network data. Node names and fingerprints shown here are examples.
An ITZA node does more than count votes.
On most networks a node orders transactions and nothing else. The application you use still runs on somebody's cloud account, which is where every real world takedown actually happens.
ITZA nodes carry the application itself. When you run one you are doing four separate jobs that would otherwise be four separate companies.
You serve the application
ITZA Press is delivered to readers straight from nodes over ordinary HTTPS. Your machine is the web host.
You deliver the content
Articles, images and audio come from the node nearest the reader. Your machine is the content delivery network.
You keep the archive
Published work is stored by its fingerprint, spread across the network. Your machine is why a piece from three years ago still loads.
You settle the money
Subscriptions, escrow releases and commission payments are confirmed by the nodes. Your machine is the payment processor.
Publishing is the case where this stops being theoretical.
An exchange going offline for an hour is an outage. A publication going offline is something people call censorship, and they are usually right.
Almost nobody who has been removed from the internet was removed by a court. They were removed by a hosting provider, a payment processor, a domain registrar or an app store, each making a commercial decision quickly and without an appeal. Every one of those is a private company with a support queue and a legal department, and every one of them can be leaned on.
Take all of them out of the path and there is nobody left to lean on. Not because anyone is defiant, but because the phone number does not exist.
This is the part that cannot be bought or engineered on its own. It is a property of how many independent people are running the thing, in how many places, under how many different jurisdictions. It is the one part of the platform we cannot build for you.
Paid for work performed, out of what the network actually earns.
ITZA does not print tokens to pay its operators. There is no inflation and no emission schedule, so nothing you receive is created out of thin air and paid for by diluting somebody else.
What operators are paid comes from protocol fees on real activity: subscriptions settling, commissions releasing from escrow, conversions on the internal exchange. If the platform is busy, that is worth something. If it is quiet, it is not. We would rather tell you that now than dress it up.
Beyond the fees
- A vote in how the protocol changes. Operators are the people who have to live with a change, so they are the people who decide on it.
- Your own copy of the archive. Not a promise from us that the work is safe. A copy, on your machine, that you can verify.
- A place in the thing itself. The platform is only as independent as the set of people running it. That is not a slogan, it is arithmetic.
Built for ordinary machines, on purpose.
ITZA runs on residential grade hardware rather than racks in a facility. That is a deliberate design choice with a specific consequence: a network of tens of thousands of modest machines in homes and offices across many countries is far harder to pressure, and far harder to quietly consolidate, than a few hundred servers in three data centres owned by two companies.
Selection of who produces blocks combines dedicated time with stake, and stake is grouped rather than ranked outright, so that simply holding the most does not let one party take over block production. Nodes that stay up and actually serve are favoured over nodes that appear only when it is profitable to appear.
What we are not going to guess at
Hardware requirements, bandwidth expectations, the stake needed, and how fees are divided are all being finalised against real testnet numbers. We will publish them as specifications rather than as a range that sounds good, and we would rather leave this section short than fill it with figures that move.
| Function | Who normally provides it | Who provides it here |
|---|---|---|
| Web hosting | A cloud provider | Every node |
| Content delivery | A commercial edge network | The node nearest the reader |
| File and media storage | Object storage at one vendor | Content addressed across the network |
| Long term archive | Backups at the same vendor | Distributed copies, each verifiable |
| Payments and payouts | A payment processor | Protocol escrow, settled by nodes |
| Deciding what stays online | Any provider above, unilaterally | Nobody, by construction |
The parts we would want to know.
- This is a job, not a yield. You are running infrastructure and being paid a share of the fees it earns. A node that is offline, or that is not serving readers, is not earning. Nothing here is an investment, and nobody is promising you a return.
- Early is quiet. Fees track usage, and a new platform has very little usage. The people who run nodes in the first year are doing it because they want the thing to exist, and we would rather say so than imply otherwise.
- It is still testnet. The network went live for testing in March 2026. Operator requirements and fee mechanics are being settled against real numbers before anyone is asked to commit hardware.
- You will be holding other people's writing. That is the point of it, and it is worth thinking about properly before you start. Storage is content addressed, which means you can verify what you hold, and the network is designed so no single operator is a chokepoint for anything.
Tell us you want to run one.
We will send the specifications, the setup guide and the fee mechanics as soon as they are settled, and nothing else in the meantime.
One email when operator details are published. Your address is never shared, sold, or used to train anything.